Date:	June 7, 2013

To:	Rachel Schmeltz, U.S. EPA

From:	Jeff Coburn, RTI International

Subject:	Uncertainty of Monthly Landfill Gas Methane Concentration
Measurements

Purpose

This memorandum documents the uncertainty of the methane recovery when
using a monthly monitoring frequency with or without a minimum
intervening period between “monthly” measurement events.

Background

On October 30, 2009, the EPA promulgated mandatory GHG reporting
requirements for a variety of sources including municipal solid waste
(MSW) landfills as 40 CFR Part 98 Subpart HH.  For MSW landfills that
have landfill gas collection, Subpart HH requires continuous monitoring
of landfill gas flow and either continuous monitoring (if continuous
monitoring equipment is present) or weekly measurement of landfill gas
methane concentration.  Industry representatives have commented that the
weekly measurement requirement is burdensome, particularly for closed
landfills, and that monthly monitoring of methane concentration would
provide an adequately accurate estimate of the methane recovery.  To
assist the EPA in their review of this issue, industry representatives
provided their weekly flow and methane concentration measurements for
the 2011 reporting year (RY2011).  The data were provided on a
landfill-specific basis with an indication of the landfill as either
active or closed.  

The EPA evaluated the weekly measurement data, as documented in the
memorandum “Review of Weekly Landfill Gas Volumetric Flow and Methane
Concentrations” from Jeff Coburn, RTI International, to Rachel
Schmeltz, U.S. EPA, dated October 18, 2012 (see Docket Item No.
EPA-HQ-OAR-2012-0934-0014).  On April 2, 2013, the EPA proposed to allow
monthly monitoring (rather than weekly monitoring) for methane
concentrations at landfills with gas collection systems based on the
results of the October 2012 data analyses.  Specifically, the proposed
amendments required that the methane concentration be determined
“…at least once each calendar month; if only one measurement is made
each calendar month, there must be at least fourteen days between
measurements.”  Industry representatives commented on the proposed
rule stating that the 14 day interval was not necessary because: (1) the
data showed little variability in methane concentrations; (2) it
“significantly and unnecessarily complicates scheduling;” and (3)
“it may not be possible to take a monthly reading at least fourteen
days apart due to the operating schedule of the device.” 

This memorandum evaluates the data originally provided by the industry
representatives in light of these comments to determine if an
intervening period between monthly samples is justified because it
serves the intended purpose of limiting the uncertainty of the methane
recovery.

Assertion 1:  The data show little variability in methane concentrations

The previous analysis (as documented in the October 18, 2012,
memorandum) used “Week 1” concentrations for each month and the
total monthly flow to develop one estimate of methane recovery.  A
second estimate of methane recovery was developed using all “Week 2”
concentration measurement.  A third and fourth estimate of methane
recovery was also developed using all “Week 3” and “Week 4”
concentration measurements, respectively.  In all of these cases, the
recovery estimates were made using concentration measurement data that
were approximately 28 days apart and had to be at least 22 days apart. 
If the rule did not include an intervening interval between monthly
samples, it is possible that the landfill owner or operator could
determine the methane concentration just before midnight on January 31,
wait a few minutes, and determine the methane concentration just after
midnight on February 1.  This would qualify as monthly samples as the
samples are taken in different calendar months, even though they are
minutes apart.  

In order to evaluate the impact of removing the intervening interval as
suggested be commenters, the original data was again evaluated using a
single methane concentration determination at the end of Month 1 for the
cumulative flow for both Months 1 and 2, the methane concentration
determination at the end of Month 3 for the cumulative flow for both
Months 3 and 4, respectively.  We note that subpart HH currently
requires weekly samples to be at least 3 days apart, therefore using
Week 4 for Month 1 and Week 1 for Month 2 would still indicate a minimum
intervening period, so this approach was not used to evaluate the impact
of removing the intervening period altogether.

The annual methane recovery was determined for each landfill based on
“no intervening period” and compared to the annual methane recovery
determined based on the weekly concentration measurements and flow
rates.  Figure 1 shows the distribution of the percent error in the
monthly measurements as determined using the “no intervening period”
values.  Figure 2 shows the same distributional analysis determined
previously using all “Week 1” methane concentration measurements. 
It is quite clear from a comparison of these figures that, while there
is no significant bias (graphs still distributed around 0 error), there
is a significantly broader range of errors.  In the true monthly
sampling paradigm (4 week intervals), there is only one landfill at the
tails (outside ±6.25 percent error).  For the “no intervening
period” analysis, there are approximately 10 landfills at each tail. 
Under the true monthly sampling paradigm, the extreme values were -6.3
percent and +6.5 percent; under the “no intervening period”
analysis, the extreme values were -21 percent and +13.2 percent.  These
results strongly indicate that more accurate methane recovery values
will be achieved if there is an intervening period between monthly
samples.

In October 18, 2012, RTI also performed a Monte Carlo analysis to
project the impacts of the variability in methane concentration
measurements on the accuracy of the calculated methane recovery value
when going from weekly sampling (n= 52 ) to monthly sampling (n=12).  In
order to project the uncertainty of monthly sampling with no intervening
period, the Monte Carlo analysis was repeated with n=6 (assuming the no
intervening period is taken to its extreme).  The 95th percentile
confidence interval for the annual average methane concentration based
on 52, 12, and 6 measurements is ±3.8%, ±7.8%, and ±11%,
respectively.  

Figure 1.  Distributional analysis of the percent difference in the
annual methane recovery using monthly methane concentration measurements
with no intervening period.

Figure 2.  Distributional analysis of the percent difference in the
annual methane recovery using monthly concentration measurements one
month apart.

Table 1 summarizes the sum of squares approach for determining the
overall uncertainty associated with the methane recovery values for a
single landfill. As seen in Table 1, the overall uncertainty of the
methane recovery values for a single reporter is expected to increase
from ±8% to ±10.5% when going from a weekly to a true monthly
monitoring frequency.  If there is no intervening period between monthly
samples, the uncertainty increases to ±13%.  Thus, the inclusion of an
intervening interval in the monthly sampling requirement helps to limit
the increase in the uncertainty associated with the methane recovery
values when switching from weekly to monthly monitoring.

Table 1.  Estimate of the Overall Uncertainty for Methane Recovery for a
Single Landfill.

Parameter	Uncertainty 	Square of Uncertainty

	N=52	N=12	N=6	N=52	N=12	N=6

Accuracy of flow meter	±5%	±5%	±5%	25	25	25

Accuracy of methane concentration	±5%	±5%	±5%	25	25	25

Variability of methane concentration	±3.8%	±7.8%	±11%	14.44	60.84	121

Sum of Squares



64.44	110.84	171

Overall Uncertainty (square root of Sum of Squares)	±8.0%	±10.5%
±13.0%



	

Assertion 2:  The intervening period “significantly and unnecessarily
complicates scheduling”

No data were presented by industry commenters as to how the intervening
period “significantly” complicates scheduling.  If the sampling is
planned for approximately the 15th of every month, the subsequent
month’s sampling can be conducted any time during the following month.
 Thus, if the landfill owner schedules sampling to be conducted near the
15th of each month, the landfill owner of operator has significant
latitude to reschedule a certain month’s sampling event if an issue
arises with the recovery system and the system cannot be sampled on the
15th.  Therefore, it is not clear how this requirement
“significantly” complicates scheduling.  Furthermore, the weekly
sampling requirement currently in subpart HH requires a 3 day interval
between weekly sampling.  As such, monthly sampling with a 14 day
intervening period is much less burdensome or complicated than the
existing weekly measurement requirement with a 3 day intervening period.
 As for “unnecessarily” complicating the rule, the “no intervening
analysis” presented under Assertion 1 provides strong evidence that
the intervening interval is necessary to limit measurement error and
uncertainty associated with monthly sampling.  

Assertion 3:  The intervening period may make it impossible to collect
samples for some months at some landfills.

The majority of landfills operate their recovery system continuously. 
From the weekly monitoring data submitted from the industry, only 22 of
395 landfills had landfill gas flow for less than 52 weeks per year and
only 12 of 395 landfills had landfill gas flow for less than 50 weeks
per year. If a landfill’s gas collection system is down for only one
or two weeks a year, the 14-day intervening period will not hinder the
collection of monthly samples.  On the other extreme, if a landfill has
no recovered gas flow for an entire month, then that landfill would have
to use missing data procedures regardless of whether or not there is an
intervening period between monthly samples (in fact, no concentration
measurement is need since the cumulative flow for the month is zero). 
If the landfill owner or operator generally plans to sample early each
month or near the 15th of the month, the next month’s sample can be
taken anytime during the month.  The only time the lack of an
intervening period would allow a landfill owner or operator to collect a
monthly gas sample is when there is only gas flow during the last week
of a month and during the first week of the next month.  The weekly flow
periods for the 12 landfills that have landfill gas flow for less than
50 weeks per year were evaluated to determine how often the 14 day
intervening period would prevent a landfill owner or operator from
collecting a monthly sample that they could have otherwise collected if
there were no intervening period between monthly samples. We could not
identify one month for any landfill within the entire data set where
missing data provisions would have to be used under the monthly sampling
with a 14-day intervening period but not under the monthly sampling with
no intervening period.  Based on the industry’s weekly sampling data
and this analysis, we conclude that the 14-day interval between monthly
sampling events will not increase the prevalence at which facilities
must rely on missing data procedures. 

  PAGE   \* MERGEFORMAT  5 

	

  PAGE   \* MERGEFORMAT  1 

